Manu Gaur @gaur_manu
used to do physics, now multiplying matrices @CMU_Robotics | prev @IIIT_Hyderabad manugaurdl.github.io New Delhi, India Joined May 2012-
Tweets2K
-
Followers538
-
Following893
-
Likes20K
it never made sense to me when ppl said we shouldn’t use lora for RL. Even if you don’t buy the “RL updates a subnetworks” take, there is no reason for a general idea like low rank updates to not work with specific training paradigms
it never made sense to me when ppl said we shouldn’t use lora for RL. Even if you don’t buy the “RL updates a subnetworks” take, there is no reason for a general idea like low rank updates to not work with specific training paradigms
So much controversy triggered by claims about how humans learn. But the deeper question is why the way we learn works. We need to understand the why to know in what way “how” matters. Nature is inspiration for AI, not prescription. The crux of the whole debate is abstraction.…
So much controversy triggered by claims about how humans learn. But the deeper question is why the way we learn works. We need to understand the why to know in what way “how” matters. Nature is inspiration for AI, not prescription. The crux of the whole debate is abstraction.…
almost there, but not quite :)
Since compute grows faster than the web, we think the future of pre-training lies in the algorithms that will best leverage ♾ compute We find simple recipes that improve the asymptote of compute scaling laws to be 5x data efficient, offering better perf w/ sufficient compute
Pointing is a great way for improving vision-language alignment. language should not be used as a crutch for the sake of generalization, but must facilitate better visual reasoning. great release from a great team!
Pointing is a great way for improving vision-language alignment. language should not be used as a crutch for the sake of generalization, but must facilitate better visual reasoning. great release from a great team!
great blog by @setlur_amrith @aviral_kumar2 on driving knowledge acquisition during training by incentivizing the model to chain existing asymmetric capabilities. basically stitching order from chaos!
stuck in Paris, speedrunning @giffmana’s recent talk on VLMs. really great stuff, especially at the end!
Great research work. The thread is a gold mine for anyone interested in understanding diffusion language modelling and how it fares with AR models!
Great research work. The thread is a gold mine for anyone interested in understanding diffusion language modelling and how it fares with AR models!
Yup. the linear layer can reconstruct using the residual stream as long as the image is scaled. It works even if you initialize siglip with random weights :
Yup. the linear layer can reconstruct using the residual stream as long as the image is scaled. It works even if you initialize siglip with random weights : https://t.co/vlmoW3sz4v
Moving beyond MCQ to tasks that evaluate free-form generation is crucial to develop systems that better understand instructions and leverage EXISTING knowledge more effectively. From my work - gemini knows the prominent point of difference (aces VQA), but fails to independently…
Moving beyond MCQ to tasks that evaluate free-form generation is crucial to develop systems that better understand instructions and leverage EXISTING knowledge more effectively. From my work - gemini knows the prominent point of difference (aces VQA), but fails to independently… https://t.co/oENP5zxcX9
"On MMMU Pro , a visual question-answering benchmark with 10 choices, we obtain 51% shortcut-accuracy without showing the image or the question" Cambrian did show language shortcuts made by MLLMs on popular VQA datasets, but shortcuts using just the multiple choices is insane!
"On MMMU Pro , a visual question-answering benchmark with 10 choices, we obtain 51% shortcut-accuracy without showing the image or the question" Cambrian did show language shortcuts made by MLLMs on popular VQA datasets, but shortcuts using just the multiple choices is insane! https://t.co/4A6rgVNzsT

Akanksha @akankshanc
2K Followers 751 Following Passionately in love with Science, mostly Altruistic, Engineer, Amateur Astronomer & Critical thinker. Current Research focus: ▫️Mechanistic Interpretability▫️
Siddarth Venkatraman @siddarthv66
608 Followers 474 Following PhD at Mila | RL and other stuff I find interesting
Varad D @varad_d33297
5 Followers 345 Following
Z0r0 @WiseMonkey44
23 Followers 588 Following
Harman @harman_hrman
0 Followers 8 Following
Jinghua Zhong @zhongjinghua
0 Followers 4K Following
Ainesh Chatterjee @ain3sh
274 Followers 1K Following I code often and sleep occasionally. I like systems. Agents are graphs. CS{ML}+Math@UMD’25
Aryaman Bahl @AryamanBahl12
8 Followers 124 Following CS Researcher, @iiit_hyderabad Learning self defence
Gabriel Sarch @GabrielSarch
695 Followers 696 Following Postdoctoral Fellow @PrincetonPLI. Ph.D. @mldcmu @cmuneurosci. Prev. @yutori_ai @MSFTResearch.
ɢʀɛǟȶK̶i̶n̶g�... @GreatKingCnut
611 Followers 3K Following But the sea came up as usual and disrespectfully drenched the king's feet and shins. I want the good ending pls, not the bad one. transhumanist, ML, RL, lmao
MarciaFaulkner @Hl3nplE42Mpeoy
31 Followers 571 Following
Robert Scoble @Scobleizer
543K Followers 24K Following The best from ML/AI community | Ex-Microsoft, Rackspace, Fast Company | Wrote eight books about the future | Silicon Valley robots, holodecks, BCIs, & startups.
Wes Simpson @SimpsonWes
49 Followers 460 Following Undergrad Student @nyuniversity • Software Engineer @ Occupi
kshitij @Kshitijjkapoor
1K Followers 994 Following schizo cyber art museum | gsoc'24 @projecthoneynet
Darklord @Darklord1093741
0 Followers 77 Following
Bhavya Agrawalla @AgrawallaBhavya
98 Followers 380 Following Research Interests - Statistics, Deep Reinforcement Learning. PhD student @CMU CS. Prev - Math and CS undergrad at MIT (2021-24) and IISc Bangalore (2020-21).
Gretel, Vega. @Rhagir90122
29 Followers 935 Following
aashi @aashiitwt
1 Followers 9 Following
Yanqing Liu @YanqingLiu83931
43 Followers 123 Following student researcher @Google; Phd student @ucsc; B.Eng. in CS @ZJU_China
thinkingbets @thinkingbets
159 Followers 738 Following ml | quant | crypto systematic asymmetry hunter
pdawg @prathamgrv
16K Followers 2K Following pre doctoral researcher @MSFTResearch || part time @TensorTonic
Rohin Manvi @rohin_manvi
518 Followers 360 Following phding @berkeley_ai, research @liquidai_, prev @stanford @stanfordailab, @meta
Dharmesh Kakadia @dharmeshkakadia
1K Followers 6K Following Building https://t.co/VcaMs28aTa to give post-training superpower to everyone. @mixtrainai Past @nuro @zoox @Microsoft @MSFTResearch
VegetaAvatar @VeGeTaX29
20 Followers 6K Following
Harman Singh @Harman26Singh
1K Followers 2K Following PhD student @berkeley_ai, Prev: Gemini @GoogleDeepMind, AI Resident @MetaAI. Creating intelligence.
Sathish @Sathishkuna1
101 Followers 2K Following Engineer .Currently building LanguageLift . #100xdevs✨
Pankaj Gupta @pankaj_ipynb
66 Followers 2K Following The English language can not fully capture the depth and complexity of my thoughts; So I'm incorporate Emojis into my work to better express myself 😉.
Rohan Choudhury @rchoudhury997
496 Followers 513 Following phd student at cmu https://t.co/pjU847PL2f
MiriamSamuel @3Wgk1F060S0S75
44 Followers 2K Following
Yuvraj Singh @YuvrajS9886
2K Followers 581 Following Ex - @turboml, @puch_ai | @iitmadras (left), @iiserkol, @UofMaryland, AIISC | YESIST '24 Finalist | LLM x RL | Building SmolHub, NeatRL |
dogs so cute that cou... @dogssaveworld
120K Followers 69K Following
Sumeet Motwani @sumeetrm
2K Followers 2K Following Research Intern@Microsoft Phi | ML PhD at Oxford, Previously CS at UC Berkeley
Social Use @socialuseai
255K Followers 9K Following Where Social meets AI: Exploring the future of connected intelligence
Dawid Kopiczko @dawkopi
68 Followers 399 Following
Siddarth Venkatraman @siddarthv66
608 Followers 474 Following PhD at Mila | RL and other stuff I find interesting
Dinghuai Zhang 张鼎... @zdhnarsil
4K Followers 2K Following Researcher at @MSFTResearch. Prev: PhD at @Mila_Quebec, intern at @Apple MLR and FAIR Labs @MetaAI, math undergraduate at @PKU1898.
davinci @leothecurious
3K Followers 760 Following teaching robots to see by day, learning from nature by night. in search of elegant solutions to the metaproblem. infinitely curious.
𝔊𝔴𝔢𝔯𝔫 @gwern
65K Followers 106 Following Internet besserwisser; pedantic, mean reply guy. 𝘞𝘢𝘵𝘢𝘴𝘩𝘪 𝘬𝘪𝘯𝘪𝘯𝘢𝘳𝘪𝘮𝘢𝘴𝘶! (Follow requests ignored due to terrible UI.)
Kaien Yang @kaien_yang
791 Followers 184 Following math + cs @ stanford; prev: google deepmind, citadel, d.e. shaw
Kallol Saha @ CoRL 20... @_ksaha
100 Followers 346 Following MSR Student @ CMU RI. I work on hybrid learning-and-planning methods for long-horizon tasks in robotics and beyond. Previously RA @ RRC, IIITH
Kevin Patrick Murphy @sirbayes
61K Followers 541 Following Research Scientist at Google DeepMind. Interested in Bayesian Machine Learning.
Harold Benoit @harold_matmul
478 Followers 282 Following Another day of being a researcher in theory but an engineer in practice | tech staff @LiquidAI_
Cody Blakeney @code_star
5K Followers 1K Following Data Dawg @datologyai | Formerly Data Research Lead @DbrxMosaicAI | Visiting Researcher @ Facebook | Ph.D | #TXSTFOOTBALL fan | https://t.co/4G6Jf3at5w
Jeremy Dohmann @jecdohmann
209 Followers 259 Following Research Scientist at @perceptroninc. Former @dbrxmosaicai, @realitylabs music: https://t.co/npRSJv5bVZ
kshitij @Kshitijjkapoor
1K Followers 994 Following schizo cyber art museum | gsoc'24 @projecthoneynet
Bhavya Agrawalla @AgrawallaBhavya
98 Followers 380 Following Research Interests - Statistics, Deep Reinforcement Learning. PhD student @CMU CS. Prev - Math and CS undergrad at MIT (2021-24) and IISc Bangalore (2020-21).
Pranjal Aggarwal ✈�... @PranjalAggarw16
482 Followers 111 Following PhD Student @LTIatCMU. research scientist intern @AIatMeta FAIR. Working on reasoning, computer-use agents and test-time compute. Prev @IITD
Matteo Pirotta @teopir
629 Followers 201 Following
Niccolo' Gentile @Niccolg92
465 Followers 195 Following PhD in ML, now AI Research Lead in 🇱🇺. Here mostly AI, including sharing paper reviews. Chess, philosophy, and a travel pic may appear. Opinions are my own.
lakshya @lakshyaag
783 Followers 392 Following AI engineering in Private Equity @BainandCompany, prev @mcgillu, edtech startup, @UnivofDelhi
rasdani @rasdani_
468 Followers 3K Following
Tairan He @TairanHe99
6K Followers 800 Following Robotics&AI PhD Student @CMU_Robotics Research Intern at @NVIDIA Prev: @MSFTResearch @sjtu1896 Emboddied AI; Humanoid; Robot Learning
Jiayi Pan @jiayi_pirate
13K Followers 2K Following 🧑🍳 Reasoning Agents @xAI | PhD on Leave @Berkeley_AI | Views Are My Own
Oğuzhan Fatih Kar @oguzhanthefatih
939 Followers 545 Following Machine Learning Researcher at @Apple. CS PhD @EPFL_en on multimodal foundation models. Previously @Google, @METU_ODTU, @aselsan.
Yuxiang Wei @YuxiangWei9
815 Followers 279 Following PhD candidate @IllinoisCDS | Researcher @AIatMeta (Meta FAIR). Code LLM training.
Charlie Snell @sea_snell
8K Followers 6K Following PhD student @berkeley_ai; research @cursor_ai; prev @GoogleDeepMind. My friend told me to tweet more. I stare at my computer a lot and make things
Gavin Guo @Zhen4good
577 Followers 474 Following Embodiment @MSL Previously @Apple Siri @MITIBMLab @MIT_CSAIL @BerkeleyPhysics Opinions Are My Own
Arnab @ArnabMondal96
2K Followers 494 Following ML Researcher @Apple | PhD @mcgillu + @Mila_Quebec | Undergrad @IITKgp | Formerly: @MSFTResearch @ServiceNowRSRCH @samsungresearch
Yanqing Liu @YanqingLiu83931
43 Followers 123 Following student researcher @Google; Phd student @ucsc; B.Eng. in CS @ZJU_China
罗杰斯 🇺🇦 @dhbrojas
153 Followers 850 Following Research Engineer 智谱 https://t.co/vrJX6VP8I0 | Advanced Computing 清华大学
Rohin Manvi @rohin_manvi
518 Followers 360 Following phding @berkeley_ai, research @liquidai_, prev @stanford @stanfordailab, @meta
pdawg @prathamgrv
16K Followers 2K Following pre doctoral researcher @MSFTResearch || part time @TensorTonic
Zhi Rui Tam @zraytam
541 Followers 329 Following Research scientist at Appier. PhD at NTU. Try to make stochastic parrot smarter through yelling tokens.
Hensen Juang @basedjensen
13K Followers 2K Following cluster janitor cum architect. member of the cleanup crew for coming machine gods
Wenhao Chai @wenhaocha1
2K Followers 2K Following Ph.D. Student @PrincetonCS. Prev @Stanford @UW @pika_labs @MSFTResearch @UofIllinois. I used to work on computer vision, but it's not all I do.
Joy Hsu @joycjhsu
3K Followers 302 Following CS PhD-ing @stanford & @knighthennessy. Studying visual reasoning, neuro-symbolic learning, and visual concepts @stanfordailab & @stanfordsvl.
tokenbender @tokenbender
10K Followers 726 Following pretrain/RL/distributed training • eXperiments lab
Nader Khalil🍊 @NaderLikeLadder
9K Followers 3K Following Director of Developer Tech @ NVIDIA, Co-founder/CEO https://t.co/GCUjRDOu73 acquired by NVIDIA • I laugh til I cry it's not the same on zoom •YC W20 | UCSB • views are my own
Keenan Crane @keenanisalive
38K Followers 485 Following Digital Geometer, Assoc. Prof. of Computer Science & Robotics @CarnegieMellon @SCSatCMU and member of the @GeomCollective. There are four lights.
Dharmesh Kakadia @dharmeshkakadia
1K Followers 6K Following Building https://t.co/VcaMs28aTa to give post-training superpower to everyone. @mixtrainai Past @nuro @zoox @Microsoft @MSFTResearch
Hao Liu @haoliuhl
5K Followers 174 Following Research Scientist at Google DeepMind, PhD from @Berkeley_AI
Yuvraj Singh @YuvrajS9886
2K Followers 581 Following Ex - @turboml, @puch_ai | @iitmadras (left), @iiserkol, @UofMaryland, AIISC | YESIST '24 Finalist | LLM x RL | Building SmolHub, NeatRL |
Pankaj Gupta @pankaj_ipynb
66 Followers 2K Following The English language can not fully capture the depth and complexity of my thoughts; So I'm incorporate Emojis into my work to better express myself 😉.